Africa: Dictatorial and DemocraticElectoral Systems since 1946∗

Matt Golder†

Leonard Wantchekon‡

August 17, 2004

Forthcoming in:Colomer, Josep. ed. 2004. Handbook of Electoral System Design.

London: Palgrave.

∗We would like to thank Michael Nest for his useful comments.†Department of Politics, New York University, 726 Broadway, 7th Floor, New York,

NY 10003; E-mail: mrg217@nyu.edu, url:homepages.nyu.edu/ ∼ mrg217‡Department of Politics, New York University, 726 Broadway, 7th Floor, New York,

NY 10003; E-mail: leonard.wantchekon@nyu.edu

1 Introduction

Elections have been an integral part of African politics since independence(Nohlen, Krennerich & Thibaut 1999, Ellis 2000). There have been 321legislative and 167 presidential elections in Africa between 1946 (or inde-pendence) and 1996. These elections have occurred in both democraticand authoritarian periods. Although elections have been relatively com-mon in Africa, very little scholarly attention has been paid to them. Thedearth of electoral studies focused on Africa compared to other regionsof the world can partly be explained by the widespread establishment ofsingle-party regimes in the 1960s. It was not until the reemergence of demo-cratic multi-party elections following the third wave of democratization inthe 1990s that African electoral studies began to grow (Bratton & van deWalle 1997, Cowen & Laakso 1997, Wiseman 1992, Barkan 1995, Sisk &Reynolds 1998, Manning 2002). This chapter represents an addition tothis growing literature by providing an overview of elections and electoralsystems in Africa. While focusing primarily on the electoral institutionsemployed in democratic elections, we also consider the role of elections inauthoritarian periods. We describe the particular electoral rules employed,investigate why they were chosen and examine their impact on African partysystems.

2 An Overview of African Elections

Authoritarian government has dominated the post-war history of indepen-dent Africa. In fact, there have only been 189 country-years of democracyin Africa compared to 1823 country-years of dictatorship between 1946 and2000. Moreover, dictatorships still outstrip the number of democracies inAfrica by a considerable margin despite the transitions to democracy thatoccurred in the early 1990s. Table 1 indicates the number of legislative andpresidential elections that have occurred in democratic and authoritarianperiods in each African country. It is interesting to note that only Eritreaand Somaliland have failed to experience a legislative or presidential elec-tion in the time period considered here. This suggests that elections haveplayed an important role in African politics during both authoritarian anddemocratic periods.

1

Insert Table 1

Much of the recent research on African elections fails to provide a consistentdefinition of which elections should be considered democratic. In fact, theelections that are treated as democratic often vary from author to author.Clearly, this is problematic if one wants to develop testable and generalizableconclusions regarding African democratic elections. In this paper, a regime isclassified as a dictatorship if the chief executive is not elected, the legislatureis not elected, there is no more than one party, or there has been no alterna-tion in power (Przeworski et al. 2000). In other words, a regime is democraticif those who govern are selected through contested elections. Countries arecoded based on the regime that existed at the end of the given year. Thiscoding of democracy correlates highly with other attempts to classify democ-racy (Bollen 1990, Coppedge & Reinicke 1990, Gastil 1990, Gurr 1990). Thecoding results in 47 legislative and 25 presidential elections being classifiedas democratic. Thus, the vast majority of elections in Africa have occurredunder dictatorship.

The number of elections considered democratic here is much lower thanthat often seen in other studies of African elections. For example, Mozaffar(2001) is not unusual in classifying as many as 91 legislative elections as be-ing democratic between 1946 and 2000. The classification used here omitselections such as those in Botswana that are typically considered democratic.The problem with the elections in Botswana is that the same party has beenruling since independence. Thus, there is some uncertainty as to whetherelections are held in Botswana only because the ruling party is certain to winthem and whether the ruling party would yield office if it ever lost. Holm(1988) notes that ‘the resulting conflict could well force the BDP [BotswanaDemocratic Party] to choose between losing in parliamentary elections andabandoning elections as a method of leadership selection. Given the pater-nalistic attitude of the BDP from President Masire down, the latter choicewould not be surprising.’ If there is no alternation in power, regimes aretreated as dictatorships. The benefit of our classification of African regimescompared to those used elsewhere is that it is consistent, stated clearly andbased entirely on observables rather than subjective judgements.

2

3 Electoral Institutions under Dictatorship

The renewed interest in African elections is obviously related to the tran-sitions to democracy and multi-party systems that occurred in the 1990s.However, it is important to remember that the vast majority of Africanelections have occurred under authoritarian rule. Thus far, we have accu-mulated little systematic knowledge concerning the role that elections playunder dictatorship. African dictators often held elections, had legislaturesand organized political parties. It remains an open research question asto why these seemingly democratic electoral institutions were chosen giventhat ‘parties do not compete, elections do not select, and legislatures donot decide’ in dictatorships (Gandhi 2004). We do not pretend to providean answer to this question here. Instead, we provide a brief historical de-scription of the electoral institutions chosen during authoritarian periodsand pinpoint some of the implicit explanations that exist in the literaturein the hope that other scholars will conduct more detailed theoretical andempirical analyses in this area.

Electoral experience on the part of Africans was largely absent during thecolonial period (Wiseman 1990). The few notable exceptions tended to befound in francophone Africa, where the ideology of cultural assimilation oc-casionally permitted African electoral participation. For example, Frenchcolonial settlements in four Senegalese communes were allowed to vote fora deputy in the French National Assembly after 1848 (Ellis 2000, Cowen &Laakso 1997). However, it was only in the final years of colonialism followingWorld War II that African electoral participation became more widespreadas the French and British attempted to mollify emerging nationalist move-ments (Ellis 2000, Nohlen, Krennerich & Thibaut 1999). The British beganestablishing fledgling parliamentary systems that would eventually form thebasis for independence. The French ordinances of 1945 established electoralcolleges by which Africans could elect representatives to the ConstituentAssembly so as to participate in the writing of the Fourth Republic’s con-stitution. This was followed by the introduction of universal suffrage anda high degree of internal autonomy in francophone Africa in the Loi Cadreof 1956. The French ultimately hoped to maintain their influence in Africaby establishing a French dominated federal community. The first Africanelections based on universal suffrage and unrestricted party formation wereheld in the late 1950s just before the formal date of independence for mostcountries (Cowen & Laakso 1997). Similar electoral reforms to those in-

3

stituted by the French and British occurred much later in the Belgian andPortuguese colonies. It is clear to see that the colonists only brought democ-racy to Africa as they left (Adejumobi 2000).

Multiple parties competed in the early elections following independence andvoters often had a considerable range of choices. For example, Wiseman(1990) states that there were 130 different parties in Somalia at one stagefollowing decolonization in 1960. However, these multi-party systems did notlast long and were soon replaced by single-party rule in the 1960s. In coun-tries such as Togo, Benin, Sierra Leone, Kenya, Zambia, Tanzania, Angolaand Sudan, one party rule was introduced de jure. Tanzania and Kenya bothused acts of parliament to make opposition parties illegal. In other Africanstates, a de facto one party system emerged after smaller parties dissolvedvoluntarily and their leaders and voters became absorbed into the rulingparties (Wanyande 2000, Adejumobi 2000). This was the case in the Gam-bia, Botswana, Mauritius, Zimbabwe and Senegal. Many African dictatorscontinued to hold periodic elections throughout these authoritarian periods.Legislatures were often retained and parliamentary systems became increas-ingly presidential in nature. A handful of countries did temporarily returnto multi-party politics (Ghana 1969, 1979, Nigeria 1979, 1983, Burkina Faso1970, 1978, Uganda 1980, Central African Republic 1981, and Sudan 1986).However, multi-party systems did not reemerge with any real frequency inAfrica until the 1990s.

African dictators attempted to justify the shift towards single-party rule inseveral ways. For example, some claimed that single-party systems weremore suitable for the nation building projects that became necessary follow-ing decolonization; multi-party politics would only lead to ethnic conflictand division (Adejumobi 2000). Others stated that single-party rule wasrequired to help economic development. This notion was supported at thetime by scholars who thought that dictatorships were necessary to gener-ate development and that ‘political participation must be held down, atleast temporarily, in order to promote economic development’ (Huntington& Nelson 1976, Galenson 1959, De Schweinitz 1959). Still others simplyclaimed that multi-party politics was foreign to Africa and was an unnatu-ral imposition by the former colonial powers (Hanna 1964). In many cases,these justifications simply hide the obvious fact that ruling elites and dicta-tors wanted to retain power and saw multi-party elections as too risky.

The question is why these dictators chose to have a single party, hold elec-

4

tions, and maintain legislatures at all; they could have simply relied on theuse of force. After all, other dictators did eschew these institutions. Forexample, Sultan Sa’id ibn Taymur al Sa’id ruled Oman for 38 years withoutany parties or legislatures (Gandhi & Przeworski 2001). A scholarly con-sensus on the answer to this question has not yet emerged. What is clearis that dictators saw elections as a tool to control the population. MiltonObote, a former Ugandan president, clearly made this point when he statedthat elections are a way of controlling the people rather than being a meansthrough which they could control him (Cohen 1983). Wanyande (2000) iscorrect in stating that single-party elections are meaningless as measuresof popularity and legitimacy. Instead, it is more useful to see elections inAfrican dictatorships as a means for recruiting the political elite or as cere-monial performances that help enforce citizen obedience, induce complicity,and socialize the electorate (Chazan 1979, Cliffe 1967). The maintenanceof legislatures allows outsiders to achieve policy compromises and share theprivileges of power, while parties act as a stable system of patronage to re-ward their own clientele and coopt political opponents (Nohlen, Krennerich& Thibaut 1999, Gandhi & Przeworski 2001). While these claims about therole of electoral institutions under dictatorship are often case-specific andanecdotal, they generate the testable hypothesis that dictatorial survivalshould be positively related to the presence of these institutions. Gandhi(2004) uses a duration model to examine the impact of elections, parties andlegislatures on the survival of 512 dictators between 1946 and 1996 in 138countries worldwide. She finds that the ideal institutional arrangement fordictatorial survival is to have a single party with periodic elections; havinga legislature also helps. It seems that in many cases, African dictators chosethe ideal institutional arrangement for maintaining their power.

The introduction of one-party rule has had several important consequencesfor the development of African politics. The most obvious is that there waslittle alternation among parties and government during the authoritarianperiod.1 In fact, there was no alternation in national government in Africauntil the electoral victory of the Mauritian socialists in 1982 and 1983. Thelack of real choice often led to low electoral participation and turnout in elec-tions during this period (Adejumobi 2000, Cohen 1983). More significantly,the absence of formal competition was often used to justify military coups.For example, the 1966 army coup in Ghana was justified on the groundsthat the army and police had used the only means available for removinga dictator; only undemocratic methods could be used in the pursuit of theultimate goal of democracy.

5

Single-party rule has also had much longer lasting consequences as well.Many ruling parties have been able to maintain their control on powerdespite the reemergence of multi-party systems in the 1990s. Oppositionparties have often been too weak and fractionalized from decades of sup-pression and intimidation to be electorally competitive. In fact, ruling par-ties have continued to alter electoral rules to maintain their hold on power(Adejumobi 2000). This helps to explain why there were opposition boycottsof some sort in 11 of the 15 founding elections that occurred between 1995and 1997 (Bratton 1998). Ruling parties have also been able to use stateresources and exploit the patronage networks that had been developed tocoopt opposition groups during the period of single-party rule. This is not tosay that there has been no alternation in power following the emergence ofmulti-party elections. After all, regime change did follow the founding elec-tions in Cape Verde, Sao Tome and Principe, Benin, Zambia, Mali, Congo(Brazzaville), Madagascar, Niger, Lesotho, Burundi, Central African Re-public, South Africa and Sierra Leone. However, second elections have notbeen marked by leadership alternation and ruling parties have been ableto retain and consolidate their grip on power (Bratton 1998, Ellis 2000).In fact, clear presidential turnover occurred in only two cases (Benin 1996,Madagascar 1997). In both cases, members of the elite in the former dic-tatorship were returned to power. Former ruling parties continue to enjoylegislative majorities in most of sub-Saharan Africa (Manning 2002). Thisis quite intriguing and leads some to wonder whether in many cases multi-party elections are simply another form of manipulation by political elitesfoisted upon reluctant incumbent regimes by donor governments and finan-cial institutions (Cowen & Laakso 1997, Adejumobi 2000). More systematicresearch is required to fully understand why some former dictatorial eliteshave been able to return to power in Africa (and Eastern Europe).

4 Electoral Institutions under Democracy

Having examined the electoral institutions employed under dictatorship, wenow turn to the choice and impact of electoral systems used for the 47legislative and 25 presidential elections that took place during democraticperiods in Africa. These elections occurred in just 20 African countries (SeeTable 1).

6

4.1 Choice of Electoral Institutions

One feature that distinguishes democratic electoral systems is whether theregime is presidential or parliamentary. Several different criteria have beenproposed for classifying these regimes (Shugart & Carey 1992, Lijphart1992). We follow a fairly minimalist definition, where a presidential regimeis one in which the government serves at the pleasure of the elected pres-ident. The president may be directly or indirectly elected; the importantfeature is that the president selects and determines the survival of the gov-ernment. A parliamentary system is one in which the government serves solong as it maintains the confidence of the legislature. A system in which thegovernment must respond both to the legislative assembly and to an electedpresident is classified as mixed. Typically, these mixed systems are charac-terized by a president who is elected for a fixed term with some executivepowers and a government that serves at the discretion of the legislature.This classification scheme follows the recommendations of Przeworski et al.(2000).

Insert Table 2

Table 2 illustrates the type of regime employed by African countries duringdemocratic periods by colonial background. The regime adopted by Africancountries on independence was primarily parliamentary in former Britishcolonies and presidential or mixed in the colonies of other countries. Manyof the former French colonies have adopted the mixed system that has char-acterized the French Fifth Republic since 1958. The only country to havea democratic parliamentary regime that was not a former British colony isCape Verde. As Table 2 illustrates, many of the parliamentary regimes inanglophone countries became presidential during the authoritarian period ofsingle-party rule. By the 1990s, presidential regimes had come to dominateAfrican democracies (Southall 1999, Wanyande 2000); the only democraticAfrican countries that were parliamentary as of 2000 were Mauritius andCape Verde. The predominance of presidentialism raises concerns aboutthe survivability of Africa’s democratic regimes given the strong empiricalevidence that parliamentary systems survive longer than presidential ones(Cheibub 2002, Mainwaring 1993, Linz 1990, Stepan & Skach 1993). Thefact that the longest surviving democracy in Africa (Mauritius) is parliamen-tary may perhaps be no coincidence. Although Botswana is not considered ademocracy in this analysis, it is interesting to note that it too has retained a

7

parliamentary regime and been able to maintain a multi-party system sinceindependence (Southall 1999).

The vast majority of African presidents are elected using absolute majorityrule. In other words, a candidate must win over 50% of the popular voteto become president. If no candidate overcomes this threshold, then thereis a runoff between the two candidates who received the most votes in thefirst round. In many cases, a second round has not been necessary with onecandidate winning an overwhelming majority. A qualified majority systemwas used in the 1996 presidential elections in Sierra Leone. In these electionsa candidate had to win 55% of the vote in order to be elected in the firstround (Nohlen, Krennerich & Thibaut 1999). The only democratic presiden-tial elections that employed plurality rule between 1946 and 2000 occurredin the Congo (1961), Malawi (1994, 1999), Nigeria (1979, 1999) and Zambia(1996). Most states that had used plurality rule to elect presidents dur-ing the authoritarian period adopted absolute majority rule following theirtransition to democracy. Only Zambia has actually switched from using anabsolute majority requirement (in 1991) to using plurality rule (in 1996).

Having focused on presidential elections, we now examine the electoral insti-tutions that characterize the democratic legislative elections in Africa. Tra-ditionally, legislative elections have been distinguished by whether they usemajoritarian or proportional electoral systems (Duverger 1954, Taagepera& Shugart 1989, Lijphart 1998). The problematic nature of this simpledichotomy has become increasingly clear over time with the emergenceof numerous countries using more complex electoral systems that employmultiple tiers and/or a combination of electoral formulas (Massicotte &Blais 1999, Shugart & Wattenberg 2001). We classify legislative electoralsystems into three main types: majoritarian, proportional, and mixed. Ma-joritarian and proportional systems employ a single electoral formula in oneor more electoral tiers. A mixed system combines a majoritarian and propor-tional formula. Table 3 illustrates the electoral system used in democraticlegislative elections in Africa by former colony.

Insert Table 3

It is clear that the particular electoral system adopted by each countryis heavily influenced by its former colonial ruler. For example, all of themajoritarian systems that employed plurality rule are former British colonies

8

except for the Congo in 1963. Mauritius is somewhat unusual for a Britishcolony since it employs plurality rule in multi-member districts and allocatesup to eight ‘best-loser’ seats in order to ensure a fair representation foreach community. The democratic elections in Comoros, the Central AfricanRepublic and Mali all employed the absolute majority runoff formula usedin Fifth Republic France. The electoral system in Mali differed slightly fromthat in France since it used multi-member districts. The former colonies ofPortugal and Italy introduced proportional systems. Cape Verde and SaoTome and Principe both adopted the d’Hondt quota that was being used inthe Portuguese metropole. All of the other proportional systems in Africaemploy the Hare quota except for South Africa, which allocates 200 seatsusing the STV-Droop quota at the national level and 200 seats using thesame formula in nine regional constituencies. Two African countries haveused mixed systems that combine majoritarian and proportional electoralformulas. Madagascar allocated 82 seats by plurality rule and a further 78by the Hare quota in 1998. Niger employed plurality rule to allocate eightseats and the Hare quota to distribute 75 additional seats in 1993 and 1995.

Why did African countries choose the electoral institutions that they did?Clearly, the answer relates to the colonial heritage of each country. How-ever, this answer is only somewhat informative. After all, electoral rulestypically represent negotiated settlements between conflicting parties overinstitutional design (Kaminski 1999).2 Mozaffar (1998) argues that anglo-phone countries employed plurality rule for the multi-party elections of the1990s because of the institutional incentive structure that had developed un-der colonial rule and that continued to exist during the authoritarian periodof single-party rule. The British had been relatively tolerant to the forma-tion of autonomous associational life compared to other colonial powers andthis led to the creation of localized associations dealing with agriculture,welfare and industry. This, combined with the introduction of plurality ruleelections at independence, encouraged the elites to develop strong links withthe constituencies based on patronage and pork-barrel servicing. Pluralityrule was maintained following the reintroduction of multi-party elections inthe 1990s because authoritarian incumbents and opposition groups expectedto be able to retain their respective local power bases. Mozaffar claims thatelectoral institutions were different in the former French colonies becauseassociational life gad typically been organized into state-sponsored peak as-sociations. This organization structure meant that there was little incentivefor elites to develop strong constituency ties. The inclusion of these as-sociations in choosing the new electoral institutions in the 1990s ‘led to a

9

strategic convergence on proportional representation.’ It is important tonote, though, that former French colonies such as the Central African Re-public, Comoros, Mali and Congo retained majoritarian electoral systems.

The conclusion that electoral systems result from strategic negotiations be-tween conflicting parties can most clearly be seen in those countries wheredemocratic multi-party elections emerged following civil war or unrest.3 Forexample, groups with conflicting interests in Namibia and South Africa wereforced to make compromises during constitutional negotiations in order toachieve transitions to democracy. Both countries eventually introduced pro-portional electoral institutions.Consider the South African case. The Na-tional Party that held economic and political power recognized that it didnot have a realistic hope of governing under a majoritarian system dueto the presence of a permanent black majority. As a result, it sought apower-sharing formula during the negotiations prior to the first democraticelection in 1994. In particular, it wanted to replace the current plurality ruleelectoral system with a proportional one that would guarantee representa-tion to the white, colored and Indian minorities that it hoped to represent(Southall 1999). In contrast, the ANC was committed to majoritarian rulethrough a one-person, one-vote electoral system in an attempt to entrenchthe power of the majority. In order to ease the transition to democracy, bothsides were ultimately forced to compromise. The National Party acceptedthat it would not have a veto in the post-election cabinet and scrapped itsinsistence that an upper house with veto power be mandated by the con-stitution. In return, the ANC agreed that the National Party would bepart of the government for five years, that cabinet decisions would be madein the spirit of consensus, that 60% of the Parliament would be needed toapprove the final constitution, and that cabinet seats would be guaranteedto minority parties that won more than 5% of the seats in the lower house(Wood 2000). Most importantly, the ANC agreed to a proportional repre-sentation electoral system.4

Table 3 suggests that electoral systems chosen by African leaders have beenrelatively stable. Majoritarian systems employing plurality rule have beenmaintained in most of anglophone Africa. Only Sierra Leone, Namibia andSouth Africa have introduced proportional formulas. There has been littleexperimentation with alternative electoral systems in other regions of Africaas well. In fact, electoral system reform has been on the table in very fewAfrican countries (Mali, Benin and South Africa). This is somewhat sur-prising if one compares African democracies to those in Eastern Europe.

10

Although the third wave of democracy swept across both regions at roughlythe same time, East European elites have consistently and frequently experi-mented with their electoral systems (Golder forthcoming, Kaminski 2002a).This might suggest that African democracies have somehow managed toreach an equilibrium in multi-party competition (Kaminski 2002b).

However, it is important to remember that some authoritarian elites thatfailed to achieve satisfactory results at the polls simply ignored the elec-tion results. For example, the military nullified the Nigerian presidentialelections of 1993. Losing parties also failed to accept the results of the elec-tions in Angola (1992), Burundi (1993) and Lesotho (1997). In Algeria,the military interrupted the electoral process prior to the second round ofthe 1991 elections when it appeared that the Islamic Salvation Front (FIS)was going to win an overall majority. These examples suggest that manyAfrican countries failed to reach an equilibrium in multi-party competition.It may be the case that Table 3 gives a false sense of stability in that itexcludes precisely those cases where the losers were unwilling to abide bythe electorate’s decision.

It may also be premature to suggest that a long-term equilibrium has beensuccessfully achieved in those countries where the losers have accepted theelection results. To accept the electoral returns suggests that losing par-ties prefer the current institutional arrangements to open opposition andattempting to overthrow the government by extra-constitutional methods(Przeworski 1991). However, it is unclear whether this situation will last inall of these countries. The specific combination of ethnically-based partieswith plurality rule and geographically-located ethnic groups that character-izes certain African countries has generated situations in which losing partiesmay have little or no chance of ever holding executive power. Winning par-ties or alliances have been elected to power with overwhelming legislativemajorities in these circumstances and the number of competitive districtsis almost non-existent in certain countries (Sisk & Reynolds 1998). For ex-ample, Frederick Chiluba’s Movement for Multiparty Democracy won 125of the 150 parliamentary seats in the Zambian elections of 1991. In effect,several African political systems have become de facto one-party states withopposition parties ‘frozen’ in permanent opposition. If the costs to rebellionever decrease or the benefits from holding office increase, it may well becomepreferable for these parties to rebel rather than fight a futile battle on theelectoral front.

11

4.2 Impact of Electoral Institutions

This section briefly examines the impact of electoral institutions on Africanparty systems. In particular, it investigates their impact on the num-ber of parties in a given country. While there is an extensive empiri-cal literature analyzing the link between various aspects of electoral sys-tems and the number of parties, this research has focused almost exclu-sively on Western Europe or OECD countries (Sartori 1976, Taagepera& Shugart 1989, Lijphart 1998). Little systematic analysis has been con-ducted on the impact of electoral institutions on the number of parties in theAfrican context.5 This is important since there is other empirical researchto suggest that the number of parties has a significant impact on stabil-ity (Mainwaring 1993), political violence (Powell 1982), and party systemextremism (Cox 1990).

The literature that focuses on electoral institutions suggests that greaterelectoral system permissiveness should be associated with a larger number ofparties. This is because voters have less incentive to vote strategically whenelectoral systems are permissive. Equally, political elites have less incentiveto withdraw in favor of better placed parties (Duverger 1954, Cox 1997).Thus, disproportional systems are likely to be associated with few partiesand proportional systems with many. The proportionality of an electoralsystem is typically measured in terms of its district magnitude (the numberof seats allocated in a constituency). Table 4 indicates the average effectivenumber of electoral and legislative parties by average district magnitude.6

Empirical research in other regions of the world suggest that dispropor-tionality also seems to vary systematically across electoral formulas. Forexample, it appears to decrease as one moves from plurality rule to absolutemajority rule to the d’Hondt quota to the Hare quota. Table 5 indicateshow the effective number of electoral and legislative parties varies with theseelectoral formulas. In each of the tables, theory would lead one to expectthat the number of parties should increase as one moves from the left-mostcolumn to the right-most column.

Insert Tables 4 and 5

It is fairly clear that there is no monotonic increase in the number of elec-toral and legislative parties as district magnitude increases. Similarly, thenumber of parties does not increase as the electoral formula becomes more

12

permissive. Thus, it seems that electoral institutions do not (yet) have thesame impact on the number of parties in Africa as they do in other regionsof the world.

There are at least two explanations for this result. Each explanation hascontrasting implications for the distinctiveness of African party systems. Itmay simply be the case that there has been insufficient time for the strategicincentives generated by electoral institutions to be fully felt. After all, thereis only a handful of countries that have held more than two consecutive elec-tions. The fact that political parties and voters have little experience withelectoral politics serves to weaken the incentive structure of the electoralsystem. Voters can only act strategically when they know who the trailingparties are and when they have public information about the preferencesand intentions of other voters (Cox 1997). This is unlikely to be the case innew democracies since voters have no past electoral performance by whichto judge the electoral viability of particular candidates. Moreover, opinionpolls are often inherently uncertain and unstable in such an environment.This uncertainty may explain why in some African countries so many op-position parties put up candidates in the founding elections. The point isthat this uncertainty should decline as the number of elections increases.This should lead to a ‘shake-down’ process in the party system by whichvoters desert parties that are not electorally viable and party elites begin tobetter coordinate their electoral strategies. Thus, one should expect to seethe smaller African parties disappear as they are absorbed by larger par-ties or merge to form more electorally attractive organizations. This line ofreasoning suggests that the strategic incentives created by electoral institu-tions will be more fully felt over time and that African party systems willeventually come to resemble those in other regions of the world. This is thesame process that seems to be occurring in the new democracies of EasternEurope.

The second explanation suggests that electoral institutions fail to have aneffect on the number of parties because African party systems are rarelynationalized.7 Electoral institutions primarily create incentives for strategicbehavior at the district level (Cox 1997, Cox 1999). Thus, one would expectto see few parties in a single-member district; the number of parties shouldbe higher in constituencies with larger district magnitudes. One would onlyexpect to see this empirical prediction at the national level if the partysystem was nationalized (Chhibber & Kollman 1998). Thus, the empiricalevidence linking electoral system permissiveness to the number of parties in

13

West European and OECD countries implicitly indicates that these partic-ular party systems are nationalized. In contrast, many African parties haveregional fiefdoms in which the party wins almost all of the seats. Considerthe elections of 1994 in Malawi where single-member districts and pluralityrule were employed. The three main political parties were overwhelminglydominant in their core regions and exceedingly weak in the remaining two.The Alliance for Democracy won every seat in the northern region, theMalawi Congress Party won 75% of the seats in the central region, andthe United Democratic Front won 95% of the votes in the southern region(Reynolds 1995b). Although there was almost exactly one electorally vi-able party in each district, there appears to be three national parties oncethese districts are aggregated up to the national level. This example helpsto explain why it is difficult to evaluate the impact of electoral institutionsby focusing solely on the number of African parties at the national level.The fact that political parties are typically based on ethnic groups that areoften geographically-concentrated helps to explain why many African partysystems are not nationalized. These party systems are unlikely to becomenationalized until ethnic mobilization stops being the most stable source ofmaterial benefit and security for most voters (Cohen 1983). This explana-tion suggests that African party politics may well remain distinct comparedto other regions of the world.

5 Conclusion

Elections in Africa have evolved considerably from the colonial period whenonly European settlers in a few countries were able to participate. The lim-ited electoral opportunities for natives that had been introduced by the colo-nial powers in an attempt to undermine and co-opt nationalist movementsbetween 1946 and 1957 were eventually expanded to include multi-partyelections with universal suffrage in the independence years (1957-65). How-ever, the democratic institutions that were introduced after independencesoon broke down under the combined weight of regional and ethnic conflict,military coups and the increased political influence of the former Commu-nist bloc. Single-party regimes became the norm across Africa and electionsresulted in a concentration of political power and limited political represen-tation. Elections during this period essentially served as a mechanism forco-opting political rivals and endowing authoritarian regimes with a public

14

sense of legitimacy (Collier 1982). It was only with the end of the Cold Warand the emergence of domestic protest movements that multi-party democ-racy was restored in several Africa countries in the 1990s. The resurgenceof democracy in francophone Africa can also be traced to the Baule summitin 1989. It was here that the Mitterrand government switched from a policyof intervention in African internal politics to one of disengagement.

While African electoral systems appear to be greatly influenced by colo-nial legacies, it would be wrong to underestimate the extent of strategicnegotiation that goes on between political elites when determining electoralsystem choice. This is most obviously the case in countries that have expe-rienced civil war, internal unrest or rising ethnic tensions. The uncertaintygenerated by the rapid return to multi-party elections in the 1990s has ledto a high degree of political fragmentation in many countries. This uncer-tainty, along with the need to build bases of electoral support quickly, hastypically resulted in a strategy of ethnic mobilization (Posner 1999). Thecombination of ethnic mobilization and geographically concentrated ethnicgroups has often led to the emergence of regional parties. This developmentis encouraged in those countries that have adopted plurality rule electoralsystems. The fact that electoral institutions have often created incentivesto place ethnic identity at the center of African politics is clearly significant.This is because there is good reason to believe that ethnic politics lies atthe heart of Africa’s growth tragedy (Easterly & Levine 1997).

15

Table 1: Legislative and Presidential Elections in Africa(Democratic Elections 1946-2000; Dictatorial Elections 1946-1996)

Country Democratic Democratic Elections Dictatorial ElectionsPeriods Legislative Presidential Legislative Presidential

Algeria Never 0 0 6 6Angola Never 0 0 3 1Benin 1991-2000 3 2 5 3Botswana Never 0 0 6 0Burkina Faso Never 0 0 4 3Burundi Never 0 0 3 2Cameroon Never 0 0 8 6Cape Verde 1991-2000 2 2 3 0CAR 1993-2000 2 2 3 2Chad Never 0 0 3 1Comoros 1990-1994 2 1 5 3Congo 1960-1962 3 2 4 0

1992-1996Djibouti Never 0 0 4 2Egypt Never 0 0 13 8Equat. Guinea Never 0 0 6 4Eritrea Never 0 0 0 0Ethiopia Never 0 0 6 0Ethiopia2 Never 0 0 1 0Gabon Never 0 0 9 4Gambia Never 0 0 6 4Ghana 1970-1971 1 1 4 3

1979-1980Guinea Never 0 0 6 4Guinea-Bissau Never 0 0 4 1Ivory Coast Never 0 0 7 8Kenya Never 0 0 7 6Lesotho Never 0 0 2 0Liberia Never 0 0 9 8Libya Never 0 0 5 0Madagascar 1993-2000 2 1 8 7Malawi 1994-2000 2 2 7 0Mali 1992-2000 2 2 5 2Mauritania Never 0 0 5 3Mauritius 1968-2000 7 0 0 0Morocco Never 0 0 5 0Mozambique Never 0 0 4 1Namibia 1990-2000 2 2 0 0Niger 1993-1995 2 1 6 4Nigeria 1960-1965 3 2 4 2

1979-19821999-2000

Rwanda Never 0 0 5 3

16

Table 1: Legislative and Presidential Elections in Africa(Democratic Elections 1946-2000; Dictatorial Elections 1946-1996)

Country Democratic Democratic Elections Dictatorial ElectionsPeriods Legislative Presidential Legislative Presidential

Sao Tome & 1991-2000 3 2 2 0PrincipeSenegal Never 0 0 7 6Seychelles Never 0 0 6 3Sierra Leone 1961-1966 2 1 5 2

1996-2000Somalia 1960-1968 2 0 2 1Somaliland Never 0 0 0 0South Africa 1994-2000 2 0 11 0Sudan 1956-1957 2 0 6 4

1965-19681986-1988

Swaziland Never 0 0 5 0Tanzania Never 0 0 7 7Togo Never 0 0 6 4Tunisia Never 0 0 10 6Uganda 1980-1984 1 3 2Zaire Never 0 0 7 4Zambia 1991-2000 2 2 7 5Zimbabwe Never 0 0 9 2

Total 47 25 284 147

Data are taken from Przeworski et al. (2000) and Golder (forthcoming). The legislativeelections of 1958 in Sudan, 1963 in Congo and 1969 in Somalia are treated as democratic.Although Przeworski et al. code these years as dictatorships, these elections actually oc-curred prior to the transition to democracy (Gandhi 2004). Ethiopia2 is Ethiopia afterEritrea’s secession.

Editor’s note: Table 1 includes several democratic elections that do not appear in theGeneral Tables. The General Tables only classify elections as democratic if they occurin countries with more than one million inhabitants and have been given a score of 3 orlower by Freedom House since 1972.

17

Tab

le2:

Pre

side

ntia

l,M

ixed

and

Par

liam

enta

ryR

egim

esin

Dem

ocra

tic

Afr

ica,

1946

-200

0

Reg

ime

Bri

tish

Fre

nch

Port

ugues

eIt

alian

Oth

er

Pre

siden

tial

Ghana

(1979-8

0)

Ben

in(1

991-2

000)

Nam

ibia

Mala

wi(1

994-2

000)

Congo

(1960-6

2)

(1990-2

000)

Nig

eria

(1979-8

2)

(1999-2

000)

Uganda

(1980-8

4)

Sie

rra

Leo

ne

(1996-2

000)

Zam

bia

(1991-2

000)

Mix

edC

AR

(1993-2

000)

Sao

Tom

eSom

alia

South

Afr

ica

Congo

(1992-9

6)

&P

rinci

pe

(1960-6

9)

(1994-2

000)

Com

oro

s(1

990-9

4)

(1991-2

000)

Madagasc

ar

(1993-2

000)

Mali

(1992-2

000)

Nig

er(1

993-9

5)

Parl

iam

enta

ryM

auri

tius

(1968-2

000)

Cape

Ver

de

Ghana

(1970-7

1)

(1991-2

000)

Nig

eria

(1960-6

5)

Sudan

(1956-5

8)

(1965-6

8)

(1986-8

8)

Sie

rra

Leo

ne

(1961-6

6)

Data

are

taken

from

Prz

ewors

kiet

al.

(2000)

and

Gold

er(f

ort

hco

min

g).

18

Table 3: Legislative Electoral System Type in Democratic Africa(Election Dates in Parentheses)

Electoral System Majoritarian Proportional Multi Mixed

British Ghana Sierra Leone(1979) (1996)Malawi(1994,1999)Mauritius(1976,1982,1983,1987,1991,1995,2000)Nigeria(1964,1979,1999)Sierra Leone(1962)Sudan(1958,1986)Uganda(1980)Zambia(1991,1996)

French CAR Benin Madagascar(1993,1998) (1991,1995,1999) (1998)Comoros Madagascar Niger(1992,1993) (1993) (1993,1995)Mali(1992,1997)Congo(1963,1992,1993)

Portuguese Cape Verde(1991,1995)Sao Tome & Principe(1991,1994,1998)

Italian Somalia(1964,1969)

Other Namibia(1994,1999)South Africa(1994,1999)

19

Table 4: Average Effective Number of Electoral and Legislative Partiesby Average District Magnitude (Number of Elections in Parentheses)

Average District Magnitude1 2-5 6-10 > 10

Electoral Parties 4.54 3.84 3.96 2.50(12) (14) (6) (5)

Legislative Parties 3.67 2.92 3.60 2.31(18) (17) (7) (5)

Table 5: Average Effective Number of Electoral and Legislative Partiesby Electoral Formula (Number of Elections in Parentheses)

Electoral FormulaPlurality Absolute Majority d’Hondt Hare

Electoral Parties 2.91 6.73 2.49 6.47(19) (4) (5) (7)

Legislative Parties 2.61 4.57 2.1 4.55(23) (8) (5) (9)

Data are taken from Golder (forthcoming).

20

Notes

1Some have argued that competitive politics simply shifted to within the party duringauthoritarian periods. For example, Hayward (1987) claimed that in ‘Kenya, Tanzania,Nigeria and Sierra Leone, surprisingly large numbers of incumbents have been defeated byvoters unhappy with their performance.’ In some estimations, as many as one half of theincumbent deputies and at least one quarter of the government’s ministers were removedin some elections in Kenya and Tanzania.

2The debate in the literature as to whether African countries should employ propor-tional representation or majoritarian rule ignores this point (Reynolds 1995b, Reynolds1995a, Barkan 1995). Particular institutions may simply not be sustainable as an equilib-rium in certain societies. Instead, it is better to begin by examining the type of electoralinstitutions that might be acceptable to the relevant parties.

3The importance of strategic negotiations in choosing electoral rules can even be seenin authoritarian periods. For example, regional leaders in Benin adopted a rotatingpresidency in 1969 in an attempt to keep a lid on rising ethnic tension. The ‘ConseilPresidentiel’ was essentially a power-sharing arrangement similar to the one introduced inthe former Yugoslavia after Tito’s death in 1980.

4A similar compromise occurred in Namibia. The 1989 elections were supposed to electa constituent assembly that would write a constitution for an independent Namibia. Theseelections were conducted under the authority of the South African administration as defacto rulers and the UN Technical Assistance Group as representatives of the internationalcommunity. The South African administration employed a proportional electoral systemin an attempt to restrict SWAPO to less than two thirds of the National Assembly. Thiswould prevent SWAPO from being able to write a constitution without the agreement ofany other parties (Southall 1999).

5Mozaffar (1998, 2001) represent important exceptions.

6The effective number of electoral parties is calculated as 1/Σv2i , where vi is the percent-

age of votes won by the ith party. The effective number of electoral parties substitutes thenumber of seats won by a party for the number of votes it won (Laakso & Taagepera 1979).Data are missing for the number of electoral parties in 10 cases: Central African Repub-lic (1993), Comoros (1993), Congo (1992,1993), Madagascar (1993,1998), Niger (1995),Somalia (1969) and Sudan (1958,1986).

7We consider a nationalized party system to be one in which the parties are relativelycompetitive throughout the country.

21

References

Adejumobi, Said. 2000. Elections in Africa: A Fading Shadow of Democracy. In Gov-ernment and Politics in Africa: A Reader, ed. Okwudiba Nnoli. Harare, Zimbabwe:AAPS Books pp. 242–261.

Barkan, Joel. 1995. “Elections in Agrarian Societies.” Journal of Democracy 6:106–116.

Bollen, Kenneth. 1990. “Issues in the Comparative Measurement of Political Democracy.”American Sociological Review 45:370–390.

Bratton, Michael. 1998. “Second Elections in Africa.” Journal of Democracy 9:51–66.

Bratton, Michael & Nicolas van de Walle. 1997. Democratic Experiments in Africa: RegimeTransitions in Comparative Perspective. New York: Cambridge University Press.

Chazan, Naomi. 1979. “African Voters at the Polls: A Re-examination of the Role ofElections in African Politics.” Journal of Commonwealth and Comparative PoliticalStudies 17:136–158.

Cheibub, Jose. 2002. “Minority Governments, Deadlock Situations, and the Survival ofPresidential Democracies.” Comparative Political Studies 35:284–312.

Chhibber, Pradeep & Ken Kollman. 1998. “Party Aggregation and the Number of Partiesin India and the United States.” American Political Science Review 92:329–342.

Cliffe, Lionel. 1967. One Party Democracy: The 1965 Tanzania General Elections. Nairobi:East African Publishing House.

Cohen, Denis. 1983. Elections and Election Studies in Africa. In Political Science inAfrica: A Critical Review, ed. Yolamu Barongo. London: Zed Books pp. 72–93.

Collier, Ruth. 1982. Regimes in Tropical Africa: Changing Forms of Supremacy, 1945-1975. Berkeley: University of California Press.

Coppedge, Michael & Wolfgang Reinicke. 1990. “Measuring Polyarchy.” Studies in Com-parative International Development 25:51–72.

Cowen, Michael & Liisa Laakso. 1997. “An Overview of Election Studies in Africa.”Journal of Modern African Studies 35:717–744.

Cox, Gary. 1990. “Centripetal and Centrifugal Incentives in Electoral Systems.” AmericanJournal of Political Science 34:903–935.

Cox, Gary. 1997. Making Votes Count: Strategic Coordination in the World’s ElectoralSystems. New York: Cambridge University Press.

Cox, Gary. 1999. “Electoral Rules and Electoral Coordination.” Annual Review of PoliticalScience 2:145–161.

De Schweinitz, Karl. 1959. “Industrialization, Labor Controls and Democracy.” EconomicDevelopment and Cultural Change 7:385–404.

Duverger, Maurice. 1954. Political Parties. New York: Wiley.

Easterly, William & Ross Levine. 1997. “Africa Growth Tragedy: Policies and EthnicDivision.” Quarterly Journal of Economics 112:1203–1250.

Ellis, S. 2000. Elections in Africa in Historical Context. In Election Observation andDemocratization in Africa, ed. Jon Abbink & Gerti Hesseling. New York: St. Martin’sPress.

22

Galenson, Walter. 1959. Labor and Economic Development. New York: Wiley.

Gandhi, Jennifer. 2004. Political Institutions under Dictatorship PhD thesis New YorkUniversity.

Gandhi, Jennifer & Adam Przeworski. 2001. “Dictatorial Institutions and the Survival ofDictators.” Paper presented at the 2001 Annual Meeting of the American PoliticalScience Association, San Francisco.

Gastil, Raymond. 1990. Freedom in the World: Political Rights and Civil Liberties. NewYork: Freedom House.

Golder, Matt. forthcoming. “Democratic Electoral Systems Around the World, 1946-2000.” Electoral Studies .

Gurr, Ted. 1990. Polity II: Political Structures and Regime Change, 1800-1986. AnnArbor, MI: Inter-University Consortium for Political and Social Research.

Hanna, William. 1964. Independent Black Africa: The Politics of Freedom. Chicago: RandMcNally.

Hayward, Fred, ed. 1987. Elections in Independent Africa. London: Westview Press.

Holm, John. 1988. Botswana: A Paternalistic Democracy. In Democracy in DevelopingCountries: Africa, ed. L. Diamond, J. Linz & S. Lipset. Boulder, CO: Lynne RienerPublishers pp. 179–216.

Huntington, Samuel & Joan Nelson. 1976. No Easy Choice: Political Participation inDeveloping Countries. Cambridge: Harvard University Press.

Kaminski, Marek. 1999. “How Communism Could Have Been Saved: Formal Analysis ofElectoral Bargaining in Poland in 1989.” Public Choice 98:83–109.

Kaminski, Marek. 2002a. “Do Parties Benefit from Electoral Manipulation? ElectoralLaws and Heresthetics in Poland, 1989-93.” Journal of Theoretical Politics 14:325–359.

Kaminski, Marek. 2002b. “General Equilibrium Model of Multi-Party Electoral Competi-tion.” Unpublished manuscript, New York University.

Laakso, Markku & Rein Taagepera. 1979. “Effective Number of Parties: A Measure withApplication to Western Europe.” Comparative Political Studies 12:3–27.

Lijphart, Arend. 1998. Electoral Systems and Party Systems: A Study of Twenty-SevenDemocracies, 1945-90. Oxford: Oxford University Press.

Lijphart, Arend, ed. 1992. Parliamentary versus Presidential Government. Oxford Uni-versity Press.

Linz, Juan. 1990. “The Perils of Presidentialism.” Journal of Democracy 1:51–69.

Mainwaring, Scott. 1993. “The Presidentialism, Multipartism, and Democracy: The Dif-ficult Combination.” Comparative Political Studies 26:198–228.

Manning, Carrie. 2002. “Assessing African Party Systems after the Third Wave.” Paperpresented at the University of North Texas Conference on Parties and Elections inNew Democracies.

Massicotte, Louis & Andre Blais. 1999. “Mixed Electoral Systems: A Conceptual andEmpirical Survey.” Electoral Studies 18:341–366.

23

Mozaffar, Shaheen. 1998. Electoral Systems and Conflict Management in Africa: ATwenty-Eight State Comparison. In Elections and Conflict Management in Africa,ed. Timothy Sisk & Andrew Reynolds. Washington D.C.: United States Institute ofPeace Press pp. 81–98.

Mozaffar, Shaheen. 2001. “Institutions, Context and Outcomes in Africa’s EmergingDemocracies.” Presented at the 2001 Annual Meeting of the American Political Sci-ence Assocation, San Francisco.

Nohlen, Dieter, Michael Krennerich & Berhnard Thibaut. 1999. Elections in Africa: AData Handbook. Oxford: Oxford University Press.

Powell, G. Bingham. 1982. Contemporary Democracies: Participation, Stability and Vio-lence. Cambridge: Harvard University Press.

Przeworski, Adam. 1991. Democracy and the Market: Political and Economic Reforms inEastern Europe and Latin America. New York: Cambridge University Press.

Przeworski, Adam, Michael E. Alvarez, Jose Antonio Cheibub & Fernando Limongi. 2000.Democracy and Development: Political Institutions and Material Well-Being in theWorld, 1950-1990. New York: Cambridge University Press.

Reynolds, Andrew. 1995a. “The Case for Proportionality.” Journal of Democracy 6:117–124.

Reynolds, Andrew. 1995b. “Constitutional Engineering in Southern Africa.” Journal ofDemocracy 6:86–99.

Sartori, Giovanni. 1976. Parties and Party Systems: A Framework of Analysis. New York:Cambridge University Press.

Shugart, Matthew & John Carey. 1992. Presidents and Assemblies: Constitutional Designand Electoral Dynamics. New York: Cambridge University Press.

Shugart, Matthew & Martin Wattenberg, eds. 2001. Mixed-Member Electoral Systems:The Best of Both Worlds? New York: Oxford University Press.

Sisk, Timothy & Andrew Reynolds, eds. 1998. Elections and Conflict Management inAfrica. Washington D.C.: United States Institute of Peace Press.

Southall, Roger. 1999. Electoral Systems and Democratization in Africa. In Voting forDemocracy: Watershed Elections in Contemporary Anglophone Africa, ed. JohnDaniel, Roger Southall & Morris Szeftel. Brookfield USA: Ashgate pp. 19–36.

Stepan, Alfred & Cindy Skach. 1993. “Constitutional Frameworks and Democratic Con-solidation: Parliamentarism versus Presidentialism.” World Politics 46:1–22.

Taagepera, Rein & Matthew Shugart. 1989. Seats and Votes: The Effects and Determi-nants of Electoral Systems. New Haven: Yale University Press.

Wanyande, Peter. 2000. Democracy and the One-Party State: The African Experience.In Government and Politics in Africa: A Reader, ed. Okwudiba Nnoli. Harare,Zimbabwe: AAPS Books pp. 107–121.

Wiseman, John. 1990. Democracy in Black Africa: Survival and Revival. New York:Paragon House Publishers.

Wiseman, John. 1992. “Early Post-Redemocratization Elections in Africa.” Electoral Stud-ies 11:279–291.

Wood, Elisabeth. 2000. Forging Democracy from Below: Insurgent Transitions in SouthAfrica and El Salvador. New York: Cambridge University Press.

24